CONTINGENCIES OF WELFARE REFORM

Transcription

1 Behavior and Social Issues, 8, (1998) Cambridge Center for Behavioral Studies CONTINGENCIES OF WELFARE REFORM John A. Nevin University of New Hampshire ABSTRACT: Federal welfare reform legislation permits considerable variation in welfare-related provisions from state to state. These may be characterized in terms of duration and distribution of basic support, contingencies that are designed to facilitate the transition from welfare to work, and contingencies that are designed to punish failures to comply with work requirements. These varying provisions are the independent variables in a massive, unplanned nationwide experiment on the effects of contingencies at a societal level; the data, which are gradually becoming available, may tell us something about the relative effectiveness of different sorts of contingencies at this level. The political drive to "end welfare as we know it" culminated in the Personal Responsibility and Work Opportunity Reconciliation Act (briefly, Welfare Reform or WR), which President Clinton signed into law in August Welfare Reform produced major structural changes in the conditions under which government agencies distribute support to the poor. The effects of WR, as understood at the end of 1997, were summarized in a special report published in The New York Times. Based on interviews and trends in caseload data, the Times article suggested that, "In the end, it might prove easier than anyone imagined to change welfare, and even harder to change people's lives." A central concern of behavioral psychology is to bring about constructive, enduring changes in how people live. Welfare Reform may be construed as an unplanned experiment on how to effect such changes on a societal scale, and, as such, demands the scientific attention of behavior analysts as well as their concern as citizens. This article attempts to characterize the major variables in the WR experiment; it is too early to analyze its results. Welfare Reform arranges a number of constraints and options for the states, which in turn arrange benefits, opportunities, and sanctions for residents who need some form of support for their economic survival. Briefly, the states receive block grants under TANF (Temporary Assistance for Needy Families) determined by their 1995 AFDC expenditures. States have a fair amount of latitude in how they AUTHOR'S NOTE: Please address all correspondence to John A. Nevin, RR2, Box 162, Vineyard Haven, MA tnevin@worldnet.att.net. 101

2 NEVIN use block-grant funds to implement various aspects of their programs. All departures from the basic provisions of WR must be supported by funds from separate state programs and approved by waivers from the federal government. The states were required to submit their plans for WR implementation to the federal government in 1997, and those plans are now public knowledge. Early speculations suggested that each state would minimize its expenditures on programs previously supported by federal funds. This could lead to a "race to the bottom" as states vie to offer the least attractive support packages, partly to save money and partly to avoid in-migration of poor people from other states. Fortunately, there is no evidence so far of an interstate race to the bottom. In fact, many states have implemented plans that are more generous in some ways than previous federal support. The states differ widely in benefit levels and eligibility, programs designed to encourage the transition from welfare to work, and sanctions for failure to meet work requirements. These variations, expressed relative to pre-wr provisions, are the "independent variables" of the WR experiment. Some relatively immediate outcomes, such as changes in caseloads or employment of those who have received welfare under pre-wr provisions, are the most obvious "dependent variables." 1 Other socially significant outcomes such as crime rates, prison populations, homelessness, marriage and divorce rates, child abuse, out-of-wedlock pregnancy, and health may also be affected by variations in the implementation of WR (e. g., see Mattaini, 1998). Some of these effects, however, may not be evident for several years, or even a generation. To begin any sort of analysis, the independent variables must be characterized at least ordinally. The Center on Hunger and Poverty at Tufts University (1998) has recently completed a detailed review of the welfare-reform plans of all 50 states plus the District of Columbia, and I will use this review for my analysis. In order to arrive at manageable summary measures, I will concentrate on the following categories and score them by collapsing across factors listed within categories. 1 In fact, these standard terms of experimental design are not quite proper. The states' plans for WR implementation the "independent variables" are responses to federal legislation, and hence dependent variables at another level. Also, program outcomes the "dependent variables" may well feed back to influence program design and budget. 102

3 CONTINGENCIES OF WELFARE REFORM 1. Welfare benefits duration and eligibility a) WR sets the lifetime cumulative duration of support at a maximum of 60 months for all but 20% of the caseload, but states may reduce this or extend it indefinitely. b) WR excludes people convicted of a drug felony and newly arrived legal immigrants with children, but states may give cash or food-stamp assistance. c) WR requires that eligible families be covered by Medicaid for 12 months, but states may extend this period for children only or for all family members. 2. Support for efforts to get off welfare a) Job training is essential for all but the most menial tasks. It was formerly provided under the JOBS program. States may, but need not, continue or expand this program. b) Child care is essential for parents who work or pursue their education. States may continue, reduce, or increase child-care subsidies and training of child-care providers. c) Close management of individual and family cases is needed to help people to make the transition from welfare to work. 3. Sanctions removal or reduction of benefits a) If TANF work requirements are not met after 24 months, states may reduce or terminate benefits. b) Likewise, states may temporarily or permanently terminate Medicaid and reduce or terminate food-stamp support. The Tufts study gave considerable detail on each of these variables, plus others that I do not consider here, and scored them relative to pre-wr levels. Its scores on specific factors related to benefits were coded as +1, 0, -1, or -2, where + 1 reflects state programs that involve added support and -1 or -2 reflect the magnitude of the reduction permitted by WR I have summed the scores relating to the benefits listed above; the resulting scale runs from +3 to -5. Factors related to support for work opportunities were coded and summed similarly, except that + 1 or +2 refer to the generosity of additional state programs; the resulting scale runs from +6 to -2. The same approach was taken for sanctions, except that all scores were negative (or 0); the resulting scale runs from 0 to -6. The summary scores for all states and Washington, DC, are presented in Table

4 NEVIN TABLE 1 Summary scores for benefit levels and eligibility, support for working, and sanctions for failing to meet work requirements under welfare provisions arranged by all 50 states and Washington, DC, based on a comprehensive review by the Center on Hunger and Poverty, Tufts University, State Benefits Support Sanctions State Benefits Support Sanctions AK MT AL NC AR ND AZ NE CA NH CO NJ CT NM DC NV DE NY FL OH GA OK HI OR IA PA ID RI IL SC IN SD KS TN KY TX LA UT MA VA MD VT ME WA MI WI MN wv MO WY MS Figure 1 presents the median scores on each variable and the scores for a few selected states. Since the enactment of WR, the median state reduced benefits, increased work support, and increased sanctions for failure to meet work requirements, all relative to pre-wr levels. However, there is a lot of variation between states, as shown in Table 1; a few examples that suggest some interesting comparisons are shown in Figure

5 CONTINGENCIES OF WELFARE REFORM Benefits -6 Figure 1. Summary scores for three variables arranged by the states under Welfare Reform: Benefit levels and eligibility, support for working, and sanctions for failure to meet work requirements. Median scores for the 50 states and DC are presented at the left; scores for a few selected states are also shown. When firm data become available, the effects of work support can be seen bycomparing GA and IL. Both have low benefits and moderate sanctions, but EL offers substantially more work support. If work support programs have their intended effect, there should be larger caseload reductions and employment increases in IL than in GA. Likewise, the effects of sanctions can be seen by comparing IL and MS. Both states have low benefit levels and above-median work support, but MS has substantially more severe sanctions for failing to meet work requirements. If sanctions have their intended effect, there should be larger caseload reductions and employment increases in MS than in IL. The perennial question of whether behavior is changed more effectively by positive inducements or by punitive threats against a baseline of comparable benefits may be addressed by comparing IL, with substantial work support and modest sanctions, and MS, with less work support and severe sanctions. However, to answer the question, the positive value of work support (e. g., access to free child care during job training) and the negative value of sanctions (e. g., 105

6 NEVIN termination of food stamps) must be measured and equated on a common, behaviorally meaningful scale, which may or may not be dollars. There is another issue that must be faced when comparing positive and negative consequences. The research literature leaves no doubt that both can be highly effective in changing a target response in individual subjects, but they may also have some side effects. In particular, punishment or the threat of punishment tends to make people avoid the situation as well as changing their behavior within the situation (e. g., Sidman, 1989). Thus, work support and sanctions might produce similar caseload reductions but for different reasons: Work support might reduce caseloads as more people are employed, and sanctions might reduce caseloads as more people drop out of the system. Another comparison is related to the popular notion of a "welfare culture" in which recipients are trapped by the benefits they receive. This notion may be evaluated by comparing CA and VT: Both have high scores for work support, and 0 for sanctions, but they differ substantially in benefit levels. A similar comparison arises for MI and MS: Both have moderate work support and severe sanctions but differ in benefit levels. The question is whether more generous benefit levels make the transition from welfare more resistant to the availability of support for seeking work or to sanctions for failure to meet work requirements. If the contingencies implied by programs of work support or sanctions were immediately and directly effective on the behavior of individual welfare recipients, we could predict the outcomes of different state programs from research on positive reinforcement and punishment with individual subjects. For example, the research literature has shown that the relative increases in responding produced by added reinforcers are smaller if a high rate of reinforcement already exists (e. g., Herrnstein, 1970). Likewise, the decreases in responding produced by punishment or the threat of punishment are smaller against a background of more frequent reinforcement (for a review see Nevin, 1979). We should therefore predict that work-support programs will be more effective in CA than in VT, and that sanctions will be more effective in MS than MI. However, as argued by Malott (1998), the contingencies are actually quite weak and remote. For example, in MS and MI welfare recipients are threatened with the loss of food stamps if thev are not gainfully employed after 24 months. But this threat is unlikely to have much effect months before the fact; instead, the daily contingencies of dealing with welfare agency caseworkers and otherwise coping with life on the edge of economic survival are likely to take precedence. Similarly, the long-term positive contingency between economic wellbeing and education or job training is unlikely to be effective in the face of competing short-term contingencies of survival (see 106

7 CONTINGENCIES OF WELFARE REFORM Ruiz, 1998). The remoteness of long-term, large-scale contingencies from the daily interactions between individuals and their environments may explain why social planning has little impact on how people live. Nevertheless, as Malott (1998) suggests, even contingencies that are relatively weak and remote may have large cumulative effects over time and across individuals. As the years go by, we should have an opportunity to assess relations between the contingencies specified by current policies and their consequences for aggregate state statistics in the short and long term. There is a substantial WR research enterprise poised to do just that, and we may learn something about how large-scale social contingencies work (or fail). Unfortunately, as an experiment, WR is flawed because the variables have not been arranged randomly with respect to pre-existing differences between the states. For example, MS had the lowest rate of AFDC spending in 1995, and consequently receives the smallest TANF block grant; this may confound comparisons with other states. Moreover, the states may specify policy and assign budgets to various programs, but the exact ways in which a given program affects its clients is likely to depend on such factors as the training and personal styles of program administrators and caseworkers. Can we reasonably assume that all such factors will be random across states? If WR had been required to pass scrutiny by an institutional IRB before implementation, it would never be approved as an experiment because of its design flaws and the very real likelihood that many individuals and families will suffer. Nevertheless, Welfare Reform is now part of our culture, and it behooves us, as behavioral psychologists and concerned citizens, to study its programs, examine its effects, and use our expertise to make its permitted variations as effective as possible in promoting constructive, durable changes in the lives of the poor. 107

8 NEVIN REFERENCES Center on Hunger and Poverty (1998). Are states improving the lives of poor families? Tufts University, Medford, MA. Herrnstein, R. J. (1970). On the law of effect. Journal of the Experimental Analysis of Behavior, 13, Malott, R. W. (1998). Performance management and welfare reform: The three-contingency model of performance management applied to welfare reform. Behavior and Social Issues, 8, Mattaini, M. A. (1998). Reforming welfare reform. Behavior and Social Issues, 8, Nevin, J. A. (1979). Reinforcement schedules and response strength. In M. D. Zeiler & P. Harzem (Eds.), Reinforcement and the organization of behavior (pp ). New York: Wiley. Ruiz, M. R. (1998). Women and welfare reform: How well can we fare without education? Behavior and Social Issues, 8, Sidman, M. (1989). Coercion and its fallout. Boston: Authors Cooperative. 108